Tanking metrics and tank comparisons
 

Recently I've started to wonder how to effectively compare to tanks. Player A vs Player B, instead of classes or specs, in other words. A practical test that could actually be used.

Many cases may be protection warrior vs protection warrior, in which case you could do most of the "testing" by just comparing their stats (I'm willing to assume player skill levels are roughly equal, but this is an assumption). Things would get quite a bit more tricky comparing warrior to paladin, warrior to druid, druid to paladin...

What methodology would be the most accurate/easiest to execute? Parsing combat logs? Looking at damage taken vs length of boss fight? Tank A took 300,000 damage over a 3.2 minute prince encounter, and Tank B took 260,000 over a 2.9 minute prince encounter, so tank B is the superior tank?

Is a 3 minute boss fight enough data to provide statistically accurate results anyways? The same tank could take an extra 40k the next week even if the fight is the same duration, based soley on luck. I am largely focused on mitigation since KTM is an accurate (though not perfect) metric for threat generation.

And finally, if this is one person's dream (mine) would I even have any chance of actually collecting data from other players who just don't care? (Ok, so you probably don't have to answer this last one...)

You can compare damage taken and threat generation and that'll give you a rough idea.

However, there are intangibles that come into this (like streakiness of damage and average time to live) ... is an avoidance warrior really the best tank if an unlucky streak drops him into the red routinely? Is a druid really the worst tank if they have 25k+ hp and take crushing blows left right and center but are never in danger of actually dying? also, threat generation ... is there such a thing as excessive threat? (if you can keep comfortably ahead of your shadow priests, locks and so forth, do you really need *more* threat?)

No. There isn't.

More threat on tank = more threat dps classes can generate = more dps

Unless you had a good warrior with thunderfury in 1.x, you always had to watch your threat. Pretty much the same in 2.x, except that TF has since been nerfed.

PS: I've even had to watch my threat on the odd occasion while healing, usually in those frequent agro-wipe or OT needs to be second in threat but generates little rage situations

I still don't get what this obsession with Thunderfury comes from really. It's probably because TF makes a mediocre tank look like a great tank and a crap tank look decent still.
We have never had one until we recruited a new warrior when we were at Patchwerk. There was never any aggro issue on anything except aggro dumpers. I once went afk on Garr at 90 % and it was all fine.
In BC it seems even easier since everyone has Salvation and Tranquil Air. DPS can go all out almost all the time and in fact HAVE to go all out in many encounters to actually beat them.

Having said that, you can't compare tanks by looking at build or gear. I don't think this is something that you can effectively test with a collection of data. With one tank you will one shot Prince every week, with another tank (even with equal gear) you will wipe 10 times in a row until you get lucky.

You can always download the Tankpoints mod to find out how gear affects your ability to take damage, but be warned that the method used to calculate it rewards evasion a lot more than it is beneficial in reality.

How do you figure?

I mean, it treats avoidance like pure mitigation in that it always acts on X% of incoming damage to reduce it (whereas its more of a 'it works or it doesn't' binary effect X% of the time) and could be seen as less valuable in that way but, over time, it works just the same.

Well by that reasoning what does TF do for a great tank? I've found people tend to play around what they can expect from their tanks, regardless of the reality people always believed I could generate more threat than other tanks and other tanks reasoned I would generate more threat than they would and honestly if we do the same thing with a TF I'd be doing more.

I'm inclined to agree with the previous poster, more threat is always better as if you really see your top DPS go to town they should be pumping out some impressive threat, while there might be some differences in gear earlier on where one has higher health pools or another has better avoidance eventually these differences will be ironed out as each tank obtains the same gear sooner or later but your play style (and I guess your awareness) is something you'll carry with you no matter how much your gear changes.

I'd like to expoud on the above- at this point in the raid game, even though most fights are threat-insensitive, I honestly can't think of one where a dps class without salv can't give you a run for your money if they really try. By extension, I'd say most classes (even with salv) are 'somewhat' threat capped instead of DPS capped, at least on a non-insignficant amount of fights. Increasing your threat on a boss by 5% can potentially equate to as much as a 5% dps increase, which is obviously phenominal. I'm not saying that every dps class is threat limited (in fact, with toys like Soulshatter and Invisbility, most are not), but when you can say 'go allout' that much earlier in the fight, I think there's a decent case to say that creases fight time.

As to the OP, seriously, I would love some sort of potential concrete analysis, but I honestlydon't know what the the combatlogs show- changing stances, gain/loss of Shield Block, ect...I'm not sure if they can really provide a comprehensive analysis of a specific tank even on limited situations. That said, if you find a way to do it (tankpointsDone?) and evaluate how a specific tank performed, I would personally find it highly interesting.

The reason is that TF made tanking a lot easier from what I have seen by grabbing area aggro that is much higher than just the standard demo shout, meaning that you could be assured of at least a few hits from secondary targets (meaning more rage, more damage shield returns generating even more threat on these mobs).This all amounts to easily having enough time to generate threat on the "other" mobs, or even not needing to at all.
It is similar to how Thunderclap now pretty much trivializes multi mob pulls in normal instances. If you spam it you don't need much else than that hold area healing aggro.
Personally I always prefer a faster weapon to unload faster/cheaper heroic strikes instead.

In most boss fights I need to keep up Demo shout and thunder clap myself, (Apparently there is something in the brain like: if dps = on, then brain = off.) and on non aggro dump fights with a reasonable boss or trash mob, there is no way dps can draw aggro, even without Salvation.

DPS has to do over 1000 dps to overaggro on a tank going all out (not including salvation or talents), and there is no raid situation where you go in without salvation, making the gap even wider.

I would say that threat is only an issue if your tank is suffering from rage deprivation, which likely happens only in content below your gear level where it should not be an issue.
In fact, if you think that threat is the cap for dps, then why is everyone constantly flasking and buffing to the max to kill mobs ?

About tankpoints: Often you will be looking for the maximum damage spike you can survive rather than total amount of damage avoided. Damage taken over time is less important than guaranteeing survival of the highest damage spike. This is encounter specific, making it even more unpractical to compare tanks.

TF was absolutely sick for agro generation when the debuffs produced extra threat. It was also the only way to get 20% melee attack speed reduction for a very long time. TF was VERY GOOD to say otherwise is ludicrous considering the benefit it provided.

There is the occassional tank in our guild that can generate enough threat to enable our dps to go all out, however there are many enocounters which have mechanics which make staying under agro thresholds non trivial. Shadow priests especially with aoe damage to their parties generate insane amounts of threat.

1000 dps is a 1000tps which most warriors I've seen can not do. Hell even feral druids pre nerf had a hard time getting to that much threat.

1000 dps at the moment is relatively easy to do with full buffs if a boss likes to stay in one place for an extended time.

There is a difference between generating 1000 tps and having to do 1000 dps to overaggro though. 800-900 tps should not be an issue for a tank on most bosses. Even spamming just sunder armor and heroic strike alone yields what 600 tps or so and those aren't exactly the most efficient moves. I guess it depends on your raid but for most of our raids, dps is capped by gear and not by aggro.

And obviously the original Thunderfury was way too sick for aggro generation, that's why it's been nerfed about 10 times.

Without Salvation yes (noone raids seriously without paladins anymore I'd hope). With Salvation it becomes 700tps needed to maintain aggro, which tanks in our not-so-hardcore guild can sustain fairly easily, at least when they're getting beaten the shit out of.

As long as a mob has sunders, I can regularly hold 1.1+k tps for extended periods of time (and i've hit 1.6k TPS for strings of 5+ seconds in full DPS gear, even though that's kinda cheating). The warriors in guild can also hit 900-1k TPS regularly and hold that for quite a while before a string of parries/whatever screws it over, but in general our warriors manage to get 850+ tps total over fights (one with TF, one with the mace from Lurker).

Getting back on Topic... I think the best way to really compare is to try each tank. Note how they die and try again, switch the tank on a nightly basis to remove the 1 encounter test mistake. Of course, your raid has to have a minimum of fate in your tank and not actively try to make him fail.

After a week you should have a pretty good idea of who you like the most as your iron man.

Little piece of life:

I've been "Tried" on 1 encounter once in the beginning of our Karazhan and died after 40sec, ppl saying left and right that i wasn't ready for this yet, when the real reason was one of the 2 healers was afk. One day the warrior wasn't there and i tanked and we haven't had much trouble clearing like we usualy do. So ppl changed their mind...

Nitpick:

DPS Warrior with Salv (-20%stance mod * -30%salv mod) @ 1000DPS = 560TPS

Actually I (slightly) disagree, because I think that this is sorta the point of the thread. I really 'feel' that there should be some absolute metric that will in fact allow us to run a mod and/or pore over combat logs and get a sense of 'what are they doing right/wrong', and I think that's would be a wonderful thing to have.

Subjecting the raid to a trial by fire could be...unfortunate.

Yeah, the first night I got a lot of 'lol pally tank get a warrior back in here' and nowadays the mages bitch if I DON"T do the pulls on trash mobs (since they can go nuts that much easier). I think a mod that measures mitigation (and possibly hooks KTM for a overall recorded fight TPS and 'string of parries or can't hit a GD button' info) might signficantly lower the journey from A to B, and help identify who's slacking/doing it wrong.